Considerations and challenges in building an end to-end microbiome workflow
0 likes119 views
The document discusses the complexities and challenges in developing an end-to-end microbiome workflow, highlighting intrinsic factors, maturity of the field, and business value. It emphasizes the need for effective metadata management and robust compute resources, alongside advancements in industrialized microbiome workflow solutions. Eagle Genomics offers tools to streamline data collaboration, improve researcher productivity, and enhance data utilization in microbiome research.
Considerations and challenges in building an end to-end microbiome workflow
- 1. data - science - life Considerations & Challenges in Building an End-to-End Microbiome Workflow
- 2. data - science - life Craig earned his Ph.D. in Microbology at the University of Warwick, he then completed his Postdoctoral research at the John Innes Centre before joining EBI and developing their bioinformatics capability. Dr McAnulla has extensive experience in developing large scale pipelines for analyzing microbiome datasets. His skills cover both software and biology and he often provides deep insights to industry experts on these topics. Presenters Raminderpal earned his PhD in the semiconductor industry where he spent a decade building advanced technologies for semiconductor modelling and data mining. In 2012, Dr Singh took the helm as the Business Development Executive for IBM Research’s genomics program, where he developed and led the execution of the go-to-market and business plan for the IBM Watson Genomics program. In addition to many published papers, Dr Singh has two books and twelve patents. Dr Craig McAnulla Senior Consultant Bioinformatics at Eagle Genomics Dr Raminderpal Singh Vice President Business Development at Eagle Genomics Please Tweet your questions @eaglegen or enter them into GoTo Webinar
- 3. Agenda 1. Considerations 2. Challenges 3. Industrialized Microbiome Workflow & Collaboration Environment 4. Deployment of a Collaboration System data - science - life
- 4. data - science - life Considerations for successful microbiome data management and analysis: 1. Intrinsic factors 2. The relatively early maturity of the field 3. Calculation of business value 1. Considerations
- 5. data - science - life Intrinsic • Microbiomes are complex and diverse biological systems • Many components – organisms, genes, pathways, metabolites etc • More DNA/genes than human genome • Taxonomic diversity • Many sources of error • Appropriate study design and metadata collection are crucial. 1. Considerations - Intrinsic
- 6. data - science - life Maturity • Changing sequencing technologies • Move from amplicon-based to shotgun • Rapidly increasing data volumes • Limited reference data • Fast-moving software development • Future data re-analysis with improved methods must be considered. 1. Considerations - Maturity
- 7. data - science - life 1. Considerations – Business value Business value • Commercial exploitation of the microbiome is in its infancy, and still at an exploratory phase. • Many research/findings remain controversial and some early claims on the gut microbiome have not been replicated. • The true potential of microbiome data is still to be tapped, and some novel experimental designs are being developed that result in maximum value being generated.
- 8. data - science - life The analysis-at-scale of microbiomics data faces several challenges: • Metadata management • Compute resources necessary for storage and analysis • Statistical techniques Metadata Management • Crucial to analysis • Often done poorly • Needs to encompass all relevant information • Check for batch effects 2. Challenges
- 9. data - science - life 2. Challenges Data Storage and Compute Resources • Shotgun metagenomics example study - 1.7 billion paired-end Illumina reads, 266.5 Gb. • Bigger data volume than the human genome at 30x coverage. • Study size is increasing How to store this? • Analysis selection critical • Analyses MUST work with these data volumes • eg Kraken, Humann2 • Compute resources • Eg Kraken, multicore machine > 100Gigs • Solution – the cloud • AWS, Azure
- 10. data - science - life Statistical Techniques • Data normalization • rarefaction, normalization, log transformation? • Statistical techniques • parametric, nonparametric? • Confusion • No correct answer • Depends on the data • function vs taxonomy • shotgun vs amplicon 2. Challenges
- 11. data - science - life • Eagle Genomics leads the industry in configuring and deploying best-in-class microbiome workflow solutions. • These run large-scale parallelized pipelines cost-efficiently, with seamless connectivity between the source data and the scientists. 3. Industrialized Microbiome Workflow & Collaboration Environment www.eaglegenomics.com “Unilever’s digital data program now processes genetic sequences twenty times faster - without incurring higher compute costs. In addition, its robust architecture supports ten times as many scientists, all working simultaneously.” Pete Keeley, eScience Technical Lead - R&D IT at Unilever
- 12. data - science - life organization 3. Industrialized Microbiome Workflow & Collaboration Environment
- 13. data - science - life MICROBIOME DATA FILES (amplicon or shotgun) METADATA: eaglecatalog (implements security model) QUERY/ FILTERING: eaglecatalog DATA SHARING CRO’s DEPOSIT DATA ELN EXPERIMENT METADATA INTEGRATION WITH OMICS (RNASeq, proteomics) FEDERATED DATA & RESULTS INTEGRATION TAXONOMY/ FUNCTION ANALYSIS REPORTS R,QIIME POPULATION COHORTS VISUALIZATION CLOUD STORAGE eaglecatalog COLLABORATIVE ENVIRONMENT FOR PARTNERS FEEDBACK TO EXPERIMENTAL DESIGN COST OF EXPERIMENT DATA COLLABORATION ARCHITECTURE Solving the metadata problem
- 14. data - science - life 4. Deployment of a Collaboration System Eagle Genomics Proposal Accelerating metadata adoption and scale up • Offering a new rapid deployment for microbiome R&D organizations looking to cost-efficiently build world-class integrated workflows. • Set up of an initial collaboration system for your team. • 1 to 4 weeks • Enabling self-load your data to the catalog – leading to a standardized and streamlined workflow, which is ready for future scale out • Managing internal data, partner data, or public data
- 15. data - science - life Current Situation Desired State Negative Consequences of Non-action Positive Business Impact, using Eagle software Data in silos, not reused Easily searchable, discoverable, accessible and intuitive portal by researchers to all data and results Significant researcher time spent in “data wrangling”. Breakage pending, based on projected bottleneck Improved time and ability to insight through radically improved productivity of researchers No correlation between experiments Quick and easy exploratory analyses across experiments and data Missed opportunities for insights. Failure to leverage data as an asset Faster identification and generation of successful product innovation and insight, leading to reduced time to market No unified view of data Data and analysis results stored in a unified catalog, with standardized & repeatable access using best-in- class technology Data (internal, external) and the associated learning not available or exploited as necessary Competitive advantage through cutting edge research, driving innovation in the product pipeline Lack of governance and data provenance Secure authenticated access with relevant data provenance of data sets; simple data sharing with vendors/collaborators Costly, inefficient and limited data sharing with vendors; risk of inappropriate data sharing Efficient compliant, documented, evidenced, and enforced data sharing process; effective (time & cost) collaboration between vendors and internal teams Data is in-effect archived Catalog integrated in the scientific workflow, used continuously and keeping pace with innovation in the industry Missed insights; silo-ed research activities, with high barriers to collaboration Systematic data sharing and re-use, driving faster unique cross-functional insights and reducing waste in spend BENEFITS
- 16. For more info, contact: Raminderpal Singh raminderpal.singh@eaglegenomics.com In Summary, Considerations & Challenges in Building an End-to-End Microbiome Workflow data - science - life